Deepgram announced on Thursday, at the inaugural Deepgram Speak event in San Francisco, that it is acquiring Daily — the real-time audio and video infrastructure company that created and maintains Pipecat, the open-source voice agent framework Deepgram calls the most widely used in the industry. Neither company disclosed the deal amount or the expected closing timeline.
Pipecat is the orchestration layer that wires a voice agent together: speech-to-text, a large language model, text-to-speech, turn detection and the real-time network transport between them. Daily open-sourced it in 2024, and the project has since grown to more than 200 integrations and almost 400 contributors. Under the acquisition, Pipecat keeps its existing BSD V2 license, and the two companies are establishing an independent Pipecat Foundation run by a steering committee representing the broader real-time AI ecosystem — including companies that compete with Deepgram.
Daily co-founder Kwindla Hultman Kramer joins Deepgram as Head of Open Source, and the Daily team moves over as well. "Keeping Pipecat open was fundamental to our decision to join Deepgram," Kramer said in the announcement. Daily's customers should see no short-term changes: Deepgram will continue to sell, support and develop all of Daily's existing products, including WebRTC, telephony and cross-platform libraries. Kramer told TechCrunch that Deepgram CEO Scott Stephenson's commitment to keeping Pipecat "open-source and vendor-neutral forever" was decisive among the acquisition interest Daily fielded.
The two companies go back a long way: both went through Y Combinator in 2016, and Deepgram's models have powered transcription inside Daily's APIs since 2019. Daily has raised more than $62 million to date from investors including Tiger Global, Nat Friedman, Lachy Groom and Kevin Weil, according to PitchBook data cited by TechCrunch. Deepgram, which reached unicorn status earlier this year with a $130 million Series C, says its platform serves more than 5,000 organizations and 650,000 developers who have processed over 200,000 years of audio. Daily runs a global real-time network of 75 points of presence.
Alongside the deal, Stephenson introduced the Real-Time Context Standard (RCS), an open protocol co-developed with the Pipecat team. RCS tackles a real gap: today's voice pipelines pass words between models but drop the audio context around them. The standard defines three metadata types — context drawn from the audio itself, such as a whisper or an unclear word; instructions for how a language model should use those signals; and guidance for how the response should sound. Adapters translate the signals for each provider's models, so an agent can respond quietly to a whisper or ask for clarification when it misheard. Deepgram's CPO Ed Anuff also previewed a Voice Agent Platform combining Deepgram's speech models with Pipecat and Daily's network.
The deal is a textbook consolidation move in a sector that has raised money at both layers: model makers like ElevenLabs, Gradium and Fish Audio, and infrastructure players like LiveKit and Vapi. By buying the orchestration framework rather than competing with it, Deepgram gets ecosystem gravity it could not have built alone — and its claim to offer "the first complete, open, production-grade platform for conversational voice AI" is company positioning, not an independent assessment. The experiment to watch is governance: a foundation steered partly by Deepgram's rivals is the strongest available guarantee of neutrality, and the strongest test of whether an acquirer can resist tilting an open project toward its own stack.
Comments (0)
Log in to join the discussion
Log InNo comments yet